On 03/28/2013 06:24 PM, Niels Möller wrote:

> Maybe one could also have a default autoincrementing nonce?


If you do that please don't make it the default. There are several cases
in DTLS where the nonce isn't simply incrementing (e.g. when receiving
packets out-of-order).

>> Does this property exist in any other nettle algorithms?
> It's (trivially) true for the block ciphers, and the gcm code separates
> key state (struct gcm_key) from message state (struct gcm_ctx). The hmac
> code does not, and I'd like to change that *if* I can find some reasonable
> way to do it.


It is pretty trivial to overcome though, i.e., just use different
contexts on each thread (you don't even need to re-initialize - just
copy the context), so I wouldn't spend resources on it unless there is
some obvious advantage which I don't see.

regards,
Nikos
_______________________________________________
nettle-bugs mailing list
[email protected]
http://lists.lysator.liu.se/mailman/listinfo/nettle-bugs

Reply via email to